🧬│ai-chat

[solar torrent]

Not quite a quick question. I just wanna make sure no member would flood the chat with W-Okada settings.

Not quite really quick. For W-Okada, go to #✨│ai-help or #1192011222023950368 and explain about your issues there.

#📰│dev-updates message

You say it like somebody gonna care about you lol.

[silk lotus]

which colab google still work for training?

[rare shoal]

Can someone tell me if I have implemented Speaker_Embeddings correctly?
https://github.com/TheDevloper2023/Tacotron2-Pipeline/tree/multi-speakers

[solar torrent]

I don't make money like that.

Also, I don't think "paid request" and "commission" are allowed in this server, as to prevent it from scamming.

However, if you gave your account to someone else, then that's your issue I guess.

[devout shale]

Is there an unlimited and free API for Gemini?

[solar torrent]

Not every chatbot service is free. Instead, there are other free LLM models you can try if you interested.

[solar torrent]

#🤖│bots message

[covert lake]

Everyone loves free things, but no one would give an unlimited free Chatbot API, they are very expensive

[drifting trellis]

yall were complaining abt kits.ai and that they only care abt money

but weight.ai is even worse

[devout shale]

I have gpt4o but I can't find Gemini2.5pro

[covert lake]

https://aistudio.google.com

it's not unlimited ofcourse though

[spare walrus]

Yall is there a working kaggle okada voice changer link??

[covert lake]

for help, please use #1192011222023950368 or #✨│ai-help :)

[swift musk]

Hey all. Just gonna drop this here for anyone that is interested in Ethical AI and AI alignment tools. These Bootstrappers clear up hallucinations and align AI with Truth, creating a Lantern and a collaborator instead of a half broken servant of deception. Give the Codex a try and see for yourself.
https://forge-of-truth.itch.io/the-aletheia-codex

[covert lake]

for sharing open source projects, please use #1359898289335566570

[covert lake]

#📰│dev-updates 👀
Took some hours adjusting like 5 guides while adding 3 other ones, gonna go eat

[hollow comet]

Are their any beatrice v2 modals i can download?

[minor blade]

none.

[grim mica]

Hi everyone, I would like to use Nano banana directly in gemini, but it tells me that it is still learning to generate images... If it can help I am in France. 👊 Thank you

[short birch]

its perfectly fine working

maybe you should use in aistudio.google.com

[hollow kernel]

Infer

[warm canopy]

I want to make my voice model

[idle nebula]

problem with vononox

when using in discord it picks up what other people say

what was the fix

[molten anvil]

Hola alguien tiene un archivo hina_modified_realtime_voice el mas actualizado?

Para usarlo desde google drive

[trail ridge]

Ai

Who is ai

[trail ridge]

Are you AI

[elder willow]

-colab

[pine acorn]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[grim mica]

Gemini says : I don't master the creation of images yet, but I can find some online.

[still sigil]

@lone moon no advertising

[lone moon]

hello sir, i am not advertising, it's just that i want some tester and i thought here i can find the interested people,

[still sigil]

it is advertising

and im telling you to stop

[mighty ether]

looking for elevenlabs alternattive, free and no limit

[covert lake]

check https://docs.aihub.gg and use help channels for help

[covert lake]

All users here are humans

[covert lake]

hina_modified_realtime is an old google colab which isn't suggested anymore, please check the docs and use help channels in english for help

[covert lake]

try asking in #1192011222023950368 , someone can surely help you there

[mighty raft]

hey

someone help me

how i can config voice changer to look like a human

[pulsar schooner]

Can you do voice to voice AI model with GOOGLE Colab ?

[minor blade]

Elaborate about your question on the #✨│ai-help

[odd hare]

Guys i have the problem that when my chat is too long it starts to lag around, does anyone know how to fix this without starting a new chat?

[short birch]

what website are you using

if aistudio

you cant fix it

because of tokens

you have

[odd hare]

ChatGpt

[short birch]

maybe the chat is long

[odd hare]

yeah its extremely long thats why its lagging is there any fix?

[short birch]

nope

no fixes at all

[odd hare]

I use it for coding and the longer the chat = more lag

xD

[short birch]

just a start new chat

[odd hare]

😢

[odd hare]

Is there an AI which doesnt have this?

[short birch]

its with html ui with deals problem

not an ai

because chatgpt and the aistudio loads all messages

[odd hare]

ah

[short birch]

all time

[odd hare]

Will it work if i clear the cache or use another browser?

[short birch]

its not deal with cache or browser

its with chatgpt site

[odd hare]

What if i remove the old messages with a tampermonkey script?

[short birch]

maybe its gonna do something

[plush mirage]

anyone who can help in n8n and meta developer call back url config??

[mortal tartan]

When people are posting models and adding audio examples why is it that those models never sound like that?

[potent cradle]

How much "faster" would a 5090 be vs a 4090 for rvc? Would it be a noticeable difference in ping/delay i can run it at

[north pelican]

yo does anyone have smth for ulquiorra from bleach

[vivid spade]

Is there a channel where I can promote my YouTube videos?

[covert lake]

Hello, we don't allow promotions, I hope you understand

[covert lake]

If you mean voice models, try looking in #1175430844685484042 :)

[north pelican]

yeah I found one, ty twin

[chilly lake]

not much faster, 5000 series is a lil weird

[north pelican]

what do I do if when I try and open the RVC gui it tells me theres no module named "soundfile"

[hot lion]

can anyone help me why my mic doesn't work when I use vac

[potent cradle]

damn that sucks

[vivid spade]

That’s fine, thank you!

[shy tangle]

hello does anyone know how to make rvc work on 50 series ?

[willow spire]

w-okada https://github.com/IllIlIlIllIl/voice-changer/releases/tag/b2335
applio https://github.com/IAHispano/Applio/issues/1020

[chilly lake]

Applio no longer needs an extra patch

all included in 3.4.0

[shy tangle]

thank you kitty

[full basalt]

yo how do i use ai to change my voice for a video

[empty flax]

yall rlly think agi could arrive by 2027 or its just all hyped up/

?

[wispy widget]

@grok is this real

[odd jungle]

Hi

[covert lake]

Hello! Wyd?

[covert lake]

gemini ate it

[covert lake]

Hello, please elaborate more in #1192011222023950368 :)

[covert lake]

Wassup, for help there's always #1192011222023950368

[full wave]

Does anyone have any good prompts for creating photos with the gemini 2.5 flash?

[covert lake]

@quaint ermine promotions aren't allowed

[covert lake]

https://ai.google.dev/gemini-api/docs/image-generation#prompt-guide

[full wave]

Thx

[solar torrent]

Not real.

[solar torrent]

There was a promoting channel here, but now it's no.

[elder willow]

wubwuh

[warm rock]

https://www.fanvue.com/aethervision

[median lantern]

there's no e girl models anymore??

[short birch]

yes, because #📜│rules

[deft ibex]

guys i have been working on a cursor project it is about to complete and i am to excited to show u guys

[short birch]

#1359898289335566570

[arctic fjord]

how to fix wokada delay?

[versed island]

hi

[covert lake]

Hello, use #1192011222023950368 for help :)

[elder willow]

colab

[covert lake]

Hello, you need any help?

[elder willow]

How to train a dataset in colab?

[haughty cave]

taking to Chatgpt way better than talking to bunch of NPC ass humans

[elder willow]

ChatGPT doesn't know how to answer this type of question

-colab

[pine acorn]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[elder willow]

-training

[covert lake]

Hello, colab shouldn't be ur first option as it's cloud, you should firstly check if your pc gpu is good enough to do it locally
I'm guessing you're talking about RVC STS Voice Models

You can check https://docs.aihub.gg for all types of guides :)

Or, use #1192011222023950368 for help

[lone hedge]

Discord's being a little silly

[hot lion]

Yeah

I tought 8m grtting hacked

[covert lake]

@hot lion @lone hedge it's a discord skill issue: https://discordstatus.com

[lone hedge]

Lol yeah

[weary temple]

@stark scarab

[stark scarab]

Wassup

[weary temple]

hacked people

[hasty ruin]

Yo is there someone I can talk to?Who can give me a role

[daring lodge]

can someone help my voice changer isnt workoing

[wind aurora]

Gm

[solar torrent]

For W-Okada, go to #✨│ai-help or #1192011222023950368 and explain about your issues there.

[solar torrent]

Good morning.

[pale crest]

Gn chat

[wispy widget]

gn

[placid zodiac]

can sm1 help me download vonovox, im confused

oh wrong chat mb

[buoyant scaffold]

hello

do you guys know so vits-svc

how to download ai voices? so i can train them

[covert lake]

So vits SVC is old, it's not used since over 2 years

RVC V2 is better, check https://docs.aihub.gg/ and #1192011222023950368

[buoyant scaffold]

yeah but i meant so i can talk to my friend in live with this voice changer

[buoyant scaffold]

sorry for the late reply

i need help

[covert lake]

will reply in https://discord.com/channels/1159260121998827560/1414977798883377292

[spark agate]

Is there a way to pause my training? So I can resume it later

@covert lake

[solar torrent]

so-vits-svc is so far old. RVC v2 is always preferred. The latter is better than former.

[polar flax]

"do you guys know plasma TVs?"

[solar torrent]

This is a wrong chat, real. For Vonovox, go to #✨│ai-help or #1192011222023950368.

[placid zodiac]

Yea mb i didnt see them the first time

[sacred niche]

what is weights?

[solar torrent]

"Weights" in "AI Hub by Weights" name is this website.

[hallow osprey]

hi, do u know guys how to automate posting comments on the pinterest under posts? I tried with zapier but pinterest aint allow to use API's

[surreal tundra]

use n8n workflow

like make one for it

[hollow hollow]

Hey guys , how can i create my voice model of character what i like ?

[tawdry pike]

use weigh

anyone elses voices lagging and cutting out

[hollow hollow]

I dont understand

[tawdry pike]

itsnthe bot but it has a website

[hollow hollow]

I have fragments of the voice that I need, can I somehow train the neural network to get a model of how it works on photographs in stable diffusion?

[tawdry pike]

think just upload a voice klip of the person

/video

[minor blade]

No.

[tawdry pike]

¨hmmm

hard to say then, just search yt then

[hollow hollow]

:///////

[minor blade]

You both go to the #✨│ai-help channel. @tawdry pike @hollow hollow

[tawdry pike]

oh my bad

[minor blade]

Dw, it's fine

[worldly stump]

hey guys!

what is the better local image generator with character consistency like Midjourney has with the Omni-Reference in V7?

[sacred niche]

hi

[barren mauve]

hello

[still sigil]

@stone bronze no advertising

[barren mauve]

hi sussy

:D

[still sigil]

ello

[barren mauve]

wats up?

[still sigil]

the ceiling

i dont wanna be at school

[barren mauve]

ooh?

[still sigil]

gimme my break

[rigid raft]

can someone help me

i downloaded the voice changer but it doesnt work

[haughty cave]

Damn chatgpt still got some more correcting to do. Cuz i got to correct it a lot

[sterile gale]

wassup

[plucky abyss]

hi

[abstract totem]

I asked that the other day

[sacred niche]

Hi

[abstract totem]

Sup

[abstract totem]

But DMs

[pale crest]

Gn chat

[gleaming summit]

кто в майнкрафт пойдёт

[supple finch]

hey

[covert lake]

Elaborate in #1192011222023950368 pls

[covert lake]

I mean it's an LLM that predicts the most probable token (vulgarly word), so yeah not perfect as of now

[covert lake]

Hello, be aware that this is an english only server :)

[covert lake]

Yeah that explains the detail of how LLMs work, @haughty cave check it out if u want

Also thx for sharing it, I was just giving a brief explaination on why it's not perfect :)

[summer pecan]

AI is still not perfect!

[lunar mist]

https://tenor.com/view/interstellar-matthew-mc-conaughey-crying-tearing-up-sadness-gif-7475281

[short birch]

(he's asking who is gonna go play minecarft with him)

[autumn lion]

For simply converting my vocals using already made RVC models (music purposes) is there a "BEST" one that is currently recommended? Local use on Windows, well specced Nvidia PC. 🙏. So far my plan was to use Appilo

[solar torrent]

Some PyTorch models (pth, ckpt) can, which one RVC voice model typically uses pth, but the safetensors file is more safe than the former.

There's a chance, but the thing is that not all "pth" files from #1175430844685484042 would be that dangerous. Most of voice model creators would unlikely to inject such the malicious code into them, unless you download a voice model from nowhere like from someone's direct message.

You're welcome.

Always make sure to download models from trusted sources. Same goes to Stable Diffusion models, but most models have go by safetensors by now.

[solar torrent]

Which one? The voice model or the program? For Applio, you can ask this in #✨│ai-help. If you meant by the best voice model you can use, you can find one in #1175430844685484042.

[autumn lion]

The program 🙂

[severe estuary]

hi

[ember obsidian]

Hey

[surreal tundra]

software name please 😄

[solar torrent]

If you refer to a screenshot, I think it's a calendar.

[full plover]

Hey guys! Im offering myself as a free web designer! Anyone that wants their website to be designed just dm me

[lament sentinel]

Hi everyone
I enjoined today
Hope to learn with you guys 🔥

[swift estuary]

How do you learn AI? I don't know where to start and I want to build a bankable skill

[smoky gust]

i accidently declined the firewall request when setting up a voice changer now everytime i go on nothing loads, how can i fix this?

[plucky abyss]

the page should better call out that this is a mobile app only.

didnt find it till i clicked try now and got a download apk lol

[covert lake]

@pine echo Please use #1359898289335566570 for your open source projects :)

[chilly lake]

go to firewall settings and delete the rule

[autumn lion]

Could you please elaborate?

[rigid raft]

i did that

can u help me

[lament sentinel]

I am an artificial intelligence student, I joined this channel to get to know creative people in this field to learn from them

[vast thicket]

Hello there

[vast thicket]

Cool, i´m software enginner student, but i like IA.

[mental prawn]

hi

[thick osprey]

Hello! What's a good AI humanizer and checker that doesn't require an account, money, or anything

just lets me use?

I'm trying to bypass ai detection and make things look normal

this is not for employment or school work. just something different

[lament sentinel]

What year are you?

[vast thicket]

First

[lament sentinel]

awesome, you’ve got plenty to learn and experience

[vast thicket]

Sure

You are veteran in IA?

[lament sentinel]

Not much, I’m still in the learning stage

[vast thicket]

Good, but you study in university or alone?

I started study python alone

[lament sentinel]

University, third year

[vast thicket]

Amazing, IA is an little dif right?

Need to know math and other things.

[lament sentinel]

Python is a beginner-friendly language, all the best 🔥

[vast thicket]

Totally

[lament sentinel]

Yes, it’s a lot of math, but it’s super fun!!

What about software engineering? how do you find it?

[vast thicket]

I imagine, when you learn it, this become less dif

[lament sentinel]

yes and with practice

[vast thicket]

software enginner is good too, but just code is not for me. In this next years i will try to find other areas like IA, ethical hacker or META. But in general software is amazing, i prefer software than hardware LOL

[lament sentinel]

true, it’s mostly coding, AI involves programming, but it focuses more on logic and analys, learning AI alongside software engineering would create a fantastic combination! 🔥

[wind token]

am i stupid or can you not download weights models

i've never used this b4

[still sigil]

notion

[blissful wagon]

pth files go to model or index i forgot

[sacred garden]

idk why but rmvpe_onnx fixes the voice cracks, was previously using crepe_tiny

[sterile gale]

hi

[solar torrent]

For W-Okada, go to #✨│ai-help and explain about your issues there.

[solar torrent]

I'm not sure what happened to "download model" option for free users. It still works for me.

[solar torrent]

For W-Okada or an RVC program, go to #✨│ai-help and explain about your usages there.

[solar torrent]

For W-Okada, explain about your issues trying to run the program in #✨│ai-help or #1192011222023950368.

[haughty cave]

Yeah and NPCs cry about it thinking it's gonna take over the world when we are very far away from that

[pale crest]

Gn chat

[deft ibex]

hello everyone , i am 14 and i am currently working on a project (Vibe coding) . AM i on right path ??

[graceful flame]

If you are looking for an AI video maker, it really depends on what you’re trying to do. Most of them have free plans or trials, but yeah, they usually come with limits like watermarks or only a few exports. If it’s just for testing or personal stuff, free is fine. But if you’re planning to use it for business purpose, Paid one is best as you get enhanced features. You can try Kapwing ,not 100 % AI but it has AI Features. its a free option if you don't mind the watermark. You can also checkout LunaBloom AI – More on the cinematic side. You throw in an idea and it builds polished, high quality clips with voiceovers. no experience required. Lumen5 – Turns text into short videos. Free plan is a bit basic, but solid for quick content.

[bold rampart]

Bruh… did you guys know that Hailuo AI announced the Start and End Frame video generation they gave will be given unlimited generations for 7 days. Get it now before it ends on 17th September

Well, now 6 days

Plus, it's free

[bold rampart]

Duh, that requires a decent GPU, at least it's online and doesn't use my GPU

[polar flax]

it's available in huggingface zerogpu

while it can be run locally using decent GPU

[bold rampart]

Sure, but I'm using a NVIDIA GeForce RTX 3050 Ti Laptop GPU with only 4GB of VRAM

[polar flax]

there's no reason to not try the zerogpu space 😔

[bold rampart]

Yeah

[polar flax]

imo Wan 2.2 is still good, and better than framepack back then

[bold rampart]

Can you give me the link tho?

[polar flax]

https://huggingface.co/spaces/zerogpu-aoti/wan2-2-fp8da-aoti-faster
https://huggingface.co/spaces/multimodalart/wan-2-2-first-last-frame

[spark agate]

anyone know how i can see which epoch is a certain point in tensor? it only tells me the step number

[surreal tundra]

thanks

[chilly lake]

tensorboard does not show epochs as they are not logged there

[spark agate]

ah i saw yeah, i looked on cmd

[chilly lake]

in the log/modelname folder there should be voice models with epoch and step numbers

[polar flax]

see the epoch & step number in saved pth files

like exactly this

[spark agate]

yup idk how i never realized that lmao

ty

[lone hedge]

Starting an #🌁│image-creations event on 9/11 is wild

[chilly lake]

in US there's 9/11 amount in gun deaths every month, other countries dgaf about 9/11

[cobalt kernel]

Anyone know what app would be suitable for a voicechanger in mac i5? Ive been trying so hard but no results. It would be great if someone also gave me a tutor

[brittle lagoon]

Can someone help me

[chrome knot]

Hi chat I’m new here i would like to know what program is used to make AI videos and photos to get started easily

[dapper drum]

Best best Karaoke modle so far?

[lyric light]

I can help you.

[dense wharf]

hey there! any server links with A.I bots?

[fallow flint]

@brittle lagoon what are you doing

[buoyant scaffold]

ill add you but ill talk later ok?

[cold spire]

hi

[queen tapir]

hi

[tawdry fossil]

hey there

[onyx bluff]

I think someone did a bs roformer karaoke model

[cobalt kernel]

Anyone know what app would be suitable for a voicechanger in mac i5? Ive been trying so hard but no results. It would be great if someone also gave me a tutor

[silver saffron]

Hey can anyone help? Ive been using llaria_rvc to convert my audios into AI generated song covers but now llaria isnt functioning well. Are there any better places that I can put my hugginface models into that will do the same thing for me without any issue

[polar flax]

please read this #📰│dev-updates message

[silver saffron]

Thank you

[icy pendant]

quick question is there a way to search what voice models are popular or high rated?

[covert lake]

Hello, you could theoretically sort models by likes in https://www.weights.com/models, even tho it would be suggested to just listen and test models

For help, please use #1192011222023950368 or #✨│ai-help next time

[meager jolt]

anybody here are available to chat?

[pale crest]

Gm chat

[green mortar]

Anyone know the largest model that I can run on my system with 4GB VRAM and 24GB RAM locally? I've tried up to 36B so far

[icy pendant]

Damn did it output at least 1 token per sec ?

[green mortar]

Yeah, it was actually fairly quick, I just need something that is really "smart" etc, I don't need speed for my use cases.

[cyan kelp]

hey, i was gone for some months or even years of this server, how does it works now?

[warped copper]

Has anyone worked with Federated Learning or has some background in it?

[tawny oracle]

what is a good software for voice cloning with your own voice, I just wanted to try it out and see if its fun

[covert lake]

Hello, be sure to check the https://docs.aihub.gg and to use the appropiate help channel :D

[tawny oracle]

mb

[ivory chasm]

how to become verified

[tall dome]

W speed❤️‍🩹

[frigid forum]

i hope ai replaces all of you

i can't wait to see it

[pale crest]

Gn chat

[solar torrent]

Better dreaming.

[cursive fractal]

Good day! I cannot apply the changer to discord.. I am able to hear the voice but it does not apply.

[still sigil]

@old berry no advertising

[molten stump]

c

[granite magnet]

Does anyone here have good experience working with DiT models and flow matching?

[lament locust]

I added a screen share option in my app, but it’s not capturing/reading my screen. Can someone help me fix this issue?

[urban horizon]

yo question

whats the best ai detection software

[fierce creek]

hello, just a question about general ai voice models -- will generating an onnx file from a model makes the voice sounds clearer?

[chilly lake]

no, it is the same model in a wrapper

[uneven scroll]

can anyone help me with the uwu girl voice changer

#🔍│find-models

[minor blade]

Go to the #✨│ai-help channel and elaborate about your issue there

[desert steppe]

hey

can u teach me how to train a voice

pelasesapelsaepl

[short birch]

no.

[short birch]

#✨│ai-help there's people that can help you if it doesnt gonna break rules

[desert steppe]

what breaks the rules?

[sage moss]

do u have time to vc

[short birch]

read the rules

duh

#📜│rules

[desert steppe]

too much bro

like do you read termmms of service of the iphone

[short birch]

there are seven parts and they are all small (not counting the TOS discord)

[sage moss]

how did u do this

@hollow tangleplsss help me

do u have time to vc

[desert steppe]

if you help me i will help u

[desert steppe]

im not a robot im a human being

[short birch]

The “Profile Board / Widgets” is still experimental / gradually rolling out to users, so not everyone has it yet.

[sage moss]

whatd u need

[sage moss]

ah okay fair

[desert steppe]

i need help

[astral matrix]

hi, someone has a notebook for colab that works for creating rvc covers like rvc cover maker but that works on the new runtime of colab?

[plucky crown]

Which are the best and most used agentic frameworks?
Frameworks keep changing. New ones come in. Old ones get outdated. I also want very fine control in the apps. I think many frameworks won't give that much control. What do you suggest?

Thirdly should I be using MCP "in combination" to the framework or whatever strategy you suggest?

do I even need openai lib? we can code raw too. But I do not know anything about it. I am new.

is MCP becoming the agentic ai standard now?

Do you recommend spending time on MCP. It is increasing in popularity and the notion of reusable decoupled microservice kind of thing via mcp
also makes sense. But let me know what you think

is MCP the future or worth spending time on

[high dirge]

why is my ai talking like 10 years after i spoke can somebody help me.

@timid musk

@polar flax pwease helpp

@weary temple @tepid basin

[weary temple]

wut

[high dirge]

help pleasee

[weary temple]

depends on what hardware u use

cuz if ur hardware is not dat good will be a lot of lag

or if its not local and is in cloud also will be lag

@high dirge dats ur answer

[high dirge]

tyyy

[weary temple]

yuh

gl

[covert lake]

Please don't ping random people and use the appriopiate channels

[covert lake]

Cloud shouldn't be ur first option, try checking the ai hub docs and the help channels to see if u can do it locally first

[covert lake]

please dont ping random people, use the right help channels and elaborate

[covert lake]

It would be better you read the rules when joining a discord server 😭

[marsh lagoon]

Were most pretrains created on applio, which has a slight error in its discriminator...?

[tepid basin]

You need more PC

[cyan phoenix]

What is this?

[polar flax]

https://tenor.com/view/patrickstar-spongebob-funnygifs-no-thisispatrick-gif-5374126

[real jungle]

it's clearly pickle jar

[polar flax]

https://tenor.com/view/bubble-bass-gif-27020256

[cobalt kernel]

Anyone know what app would be suitable for a voicechanger in mac i5? Ive been trying so hard but no results. It would be great if someone also gave me a tutor

[desert steppe]

Wdym

[desert steppe]

Idk

Sorry

[cedar osprey]

Yoo

[polar flax]

feel free to leave this server if you don't know what you're doing, or wait till you get some warn for breaking some rules

[solar torrent]

For W-Okada, go to #✨│ai-help or #1192011222023950368 and explain about your issues there. Don't try to ping other users for like this.

[solar torrent]

"How to train a voice model" might be too much for you for sure. Because not only you just ask about it in a non-help channel, but you also got lost trying to get others to understand you.

For how to train a voice model or other AIs, you either go to #✨│ai-help or #1192011222023950368 and explain about your needs there. Otherwise, asking like this in this channel won't give you any good idea, and these people would just mess you up.

[solar torrent]

Not everything is free.

[keen mountain]

it's free for student for a limited time ig, comet browser is pretty cool tho

gets most of my boring work done so i can focus on important stuff

[warm prism]

is this the area where I ask for help?

[minor blade]

Nope, you ask for help in the #✨│ai-help channel and elaborate about your issue if it's regarding RVC or realtime.

[oblique sinew]

i think i found a possible great way to easily test an LLM in accuracy, math, reliablity, stability of deep thinking... that sort of thing.

You tell it to without tools, find the square root of a very large number you provide it, accurate to as many decimal points as possible...

it will go about the task manually using the newton method and such, and the LLM that got closest is clearly supeiror

[true sonnet]

yo guys im starting a new scholarship and can you tell me wich tool (ai or not) will be the most helpfull i tried notion but i not fan (or just bad lol) thx !

[covert lake]

I don't really use specific ai tools for helping me to study currently, but I recently heard of this Open Source Project that helps with notes and quizziz https://github.com/CaviraOSS/neuropilot

[true sonnet]

thx man ! Whats different from chat gpt ? like it look the same ? are the responses differents ?

do you have a more known tool ?

i used notebook lm but its sometime limited cuz the ai of google sucks

[covert lake]

It's open source, free and costumizable, it uses different ai models providers and settings

it isn't a cloud website

[true sonnet]

how can i acces to it ? it dont work 🙁

@covert lake

[covert lake]

For AI Program help would be better to use #1192011222023950368 or #✨│ai-help
But that's a program I only heard of from a friend, I didn't actually use it since I don't usually need ai for help with school notes

be aware that open source programs are usually harder than sites like chatgpt and gemini for the average user

[true sonnet]

ok thx dude !

[covert lake]

@gentle trench Have you heard Applio added Realtime?

[gentle trench]

I have not, how is it?

if it's not as good as vonovox I probably won't use it lol

[covert lake]

I didn't really test it much, but it got improvements like vonovox, like swift pitch extraction method and spin-v2

[gentle trench]

interesting

has anyone compared the two yet?

[covert lake]

Dunno, it's not even been 24 hours since it has released #📰│dev-updates message

I mean it's the only other rvc ai realtime voice changer that is in active development

[gentle trench]

that sounds neat

[novel siren]

I downloaded Stability Manager/have SD and all that

And this shit is generating nothing but weeb pics, wtf

Nothing realistic, just anime pic after anime pic

[gentle trench]

does it work like Vonovox as in being a separate app on your pc or is it like deiteris and exists in your browser

[covert lake]

might be bc of the model you chose and the way you prompted?

[covert lake]

web user interface, just update your applio to 3.5.0 and you will see a realtime tab now

[novel siren]

I'm new to this whole thing but https://civitai.com/models/1731469/life-is-strange-alex-chen is one of the ones I downloaded

I see recommended weight 0.8 but there's no "weight" option

[gentle trench]

I use applio in Kaggle, I have no idea on how to exactly update it

[covert lake]

that's a LoRA (Low-rank adaptation) for a specific character,

LoRAs basically work as adjusting the style of a bigger model to something specific like a art style or character

The base model of this LoRA seems to be Illustrious as you can see in the details, which is mostly an anime model based on Stable Diffusion iirc

If you want realistic things, maybe try Flux.1 (either dev or schnell depending on ur local power) with some realistic loras

[novel siren]

I'm new to this whole thing so all that is foreign to me lol

[chilly lake]

Flux.1 is not quite realistic with plastic-smooth skin

[novel siren]

I had to fight just to get shit to work

had to download this flux.1, and this clip_l, and this t5xxl, and some diffusion_pytorch model

just to get ANYTHING to generate, it was error after error

[covert lake]

I remember there are some quite decent loras like https://huggingface.co/XLabs-AI/flux-RealismLora

not much aware of others tho tbh, been a while since I played with comfyui

I remember you talking about some other models recently with me, i forgot their names 😭

[chilly lake]

quote honestly, I get much better results with SDXL realistic models

[covert lake]

interesting, maybe i should try SDXL

[chilly lake]

cyberrealisticPony_v125, cyberrealisticXL_v60, epicrealismXL_vxviiCrystalclear, ponyRealism_V23ULTRA, realismIllustriousBy_v45FP16

there are many more

[covert lake]

@novel siren could be helpful to you

[novel siren]

where do i find all this stuff at

[chilly lake]

civitai

make sure you have configured settings to show nsfw stuff as well, most of it will be hidden

[novel siren]

yeah i just fixed that

I see pruned model 16 and 32

[chilly lake]

fp16

[covert lake]

the applio kaggle notebook got updated just 20 minutes ago https://github.com/IAHispano/Applio/commit/3ac68baf1e232caa2d95ff6faa61275495feac13

[novel siren]

I hope to give a custom voice ai assistant another go at some point but eh

programming makes my brain hurt

[covert lake]

feel free to use #1192011222023950368 or #✨│ai-help for AI Help

[novel siren]

People were recommending things like Kokoro or w/e and it just got more and more complicated

[gentle trench]

how do I import this into kaggle, I'm not the smartest

[covert lake]

The same way you did the initial setup iirc https://docs.aihub.gg/rvc/cloud/applio-kaggle/#notebook-creation--setup

[novel siren]

RuntimeError: Error(s) in loading state_dict for IntegratedAutoencoderKL: size mismatch for encoder.conv_out.weight: copying a param with shape torch.Size([32, 512, 3, 3]) from checkpoint, the shape in current model is torch.Size([8, 512, 3, 3]). size mismatch for encoder.conv_out.bias: copying a param with shape torch.Size([32]) from checkpoint, the shape in current model is torch.Size([8]). size mismatch for decoder.conv_in.weight: copying a param with shape torch.Size([512, 16, 3, 3]) from checkpoint, the shape in current model is torch.Size([512, 4, 3, 3]).

This after downloading realisticpony

[gentle trench]

this yes?

the additional code at the top idk if it's needed still

used to give me an error

without it

[covert lake]

yes

[covert lake]

it's already added in the notebook itself

[gentle trench]

is there a new link?

or the current importing way just adds that code

[chilly lake]

incompatible lora/embedding?

[gentle trench]

this right? I see the number of KB changed

[covert lake]

yup

EDIT: it turns out, for the way applio realtime is coded, it currently can't recognize audio devices for cloud users as I talked with Noobies

[novel siren]

Lord only knows, this is all foreign language to me currently

[fallow steppe]

anyone have cursor ai

[gentle trench]

?

[covert lake]

I think he's talking about https://cursor.com/

[covert lake]

I honestly don't use it, not sure if someone else does here,
Is there a specific reason you're asking for that? :)

[fallow steppe]

i need help purchasing it for indian account

[gentle trench]

never heard of it

@green violet I made a teto ai u might be blind https://discord.com/channels/1159260121998827560/1411557807077199914

[green violet]

yes im blind ❤️

I use glasses

[gentle trench]

https://cdn.discordapp.com/attachments/1372070062919520307/1414366391951294654/kasane-teto-png.gif

that's ok

[green violet]

we love teto

https://tenor.com/view/kasane-teto-teto-kasane-pearto-triple-baka-triple-baka-teto-dance-gif-14219668400298611625

[gentle trench]

yess

[covert lake]

https://www.youtube.com/watch?v=yjBuru8lHLs

[short birch]

дива 💜

[green violet]

Wut

[short birch]

no promotions

[mortal pivot]

🔥

[short birch]

@stark scarab @polar flax

[formal berry]

Hello

[gentle trench]

who is we

what are you

[gentle trench]

yum

[green violet]

wut

[green violet]

yummy

[weary temple]

@stark scarab

[chilly lake]

typical scam of listing properties you dont own, collecting money and disappearing

[gentle trench]

kill them

scammers go to hell

[pale swift]

-colab

[pine acorn]

📒 Google Colab Notebooks

Google Colab is a Cloud (Remote Good PC) Service. While the Free plan provides up to 12 hours of daily usage, the GPU is typically available for only about 4 hours each day on average.

• **Applio**

by IA Hispano
Google Colab

• **RVC Mainline**

by Hina
Google Colab

• **UVR5 NO UI**

by Eddy
Google Colab

• **UVR5 UI**

by Eddy
Google Colab

• **Wokada Deiteris Fork**

by Deiteris & Hina
Google Colab

• **Hina's Modified Original Wokada**

Google Colab

• **RVC-AI-Cover-Maker-WebUI**

by Shiro & Eddy
Google Colab

• **FaceFusion UI**

by Nick088
Google Colab

• **FaceFusion NO UI**

by Nick088
Google Colab

• **Music Source Separation Training (Inference)**

by Jarredou & Makidanye
Google Colab

[polar flax]

it's all about ponzi scheme, no matter how sophisticated they are

[scarlet girder]

la creatura

[timber cradle]

Hello does anyone know how to train an ai? i need some help on that

ig i should not come and ask like that 💀

but i need help fr fr

[covert lake]

Welcome here, you can get AI Help in #1192011222023950368 or #✨│ai-help

[polar flax]

go to the said help channel, but there are many types of AI so the answer would be different

[balmy mist]

Hey guys! I hope everyone is doing well.

I was wondering, which UVR5 models should I download to seperate vocals and intrumentals? I need the vocals mainly.

also which UVR5 models to get for extract and separate backing vocals as well?

[vague whale]

could any one help me get a student email ?

[queen kernel]

vocals fv4
de-reverb by anvuew
denoise-aufr33

[queen kernel]

For Gemini?

[shrewd marlin]

can somone help me how do i download rvc models exactly

[vague whale]

yes

[queen kernel]

#🔍│find-models or weights.gg

[queen kernel]

Okay. I claimed it already

[shrewd marlin]

i did find model now what

there isnt download option or smth like it

[queen kernel]

/find

[hidden grotto]

No query provided

[shrewd marlin]

ye i did

[queen kernel]

So if you go over the Weights web you will find download button

[shrewd marlin]

ok i will try this one

because it was taking me to create with ai website

[queen kernel]

Use View model button

[vague whale]

could any body tell me how can i get access to Gemini for free ex. studens email or something like that

[desert lion]

Check the official “Gemini for Students” offer — Google is running a student program that gives eligible university students Google AI Pro free for a year (or extended months). Check eligibility and sign up on the official page and follow the enrollment steps

[vague whale]

my universitry did't give the students edu mail that's why i am not able to sign in do you know how to get a edu mail or help me get it please ?

[desert lion]

Ask Your University IT/Admin
Some universities don’t give .edu emails by default but you can request them it for academic purposes (like software access, research, or internships). You can write to your IT department or registrar and ask if they can provide one.

[vague whale]

I am in Russia it's restricted country we have sanctions

[desert lion]

🙀

Idk then 👍

[formal plume]

-rvc

[pine acorn]

📚 RVC Documentations

AI HUB Docs

https://docs.aihub.gg

🍏 Applio Docs

https://docs.applio.org

[craggy jay]

is there anyway to stop the cuttin out of the voice?

[solar torrent]

For W-Okada, go to #✨│ai-help or #1192011222023950368 and explain about your issues there.

[opaque monolith]

I am a developer for a small company and I am looking for a place to talk about ai and for people to share their opinions, views and experiences regarding ai. Am I in the right place?

[grave gate]

Is it okay to ask for voice/model recommendations here?

[solar torrent]

Not really any voice model recommendation. For voice models, there are in #1175430844685484042.

[grave gate]

I meant like "Hey I wanted a female voice but with a deep pitch, anyone has something" stuff like that

But all fine

[slow palm]

Anyone replaced a Asus laptop battery?

Was wondering if it's possible with just screw driver so I don't have to ask a specialist

[plucky abyss]

start unscrewing screws until you can either get toi the battery or no more screws can be removed lol

normally laptops usualy should be semi-servicable

[slow palm]

Thanks!

[plucky abyss]

unless its one of those super thin ones

[slow palm]

I just wanna make sure that I don't break anything when replacing my battery

[slow palm]

My laptop is pretty big lol

[graceful flame]

Hi , You can try out some ai video generators to create AI videos. There are many available now, You can Just give the script/Text, it will covert them into Videos in minutes . You can try Descript if your just experimenting . its combines video editing, transcription, and podcast editing. Next one is LunaBloom AI - It helps to create High quality cinematic video based on your text. custom AI avatar also available , with natural AI voices. if its for business purpose , or for marketing , Its a best option , and you can try Fliki also this one transform text and audio to good quality videos. Heygen is a good option for image to video creation.. a quick process.

[shrewd marlin]

after i finish setup in von do i just open start bat?

[gentle trench]

yup

[magic totem]

Make video in twins development

[gentle trench]

What

[kind root]

hey guys
has anyone taken discrete structures here?
Does anyone have project ideas that use discrete structures, must be programmed in Python, and should solve real-world problems or have practical applications?

[rare nymph]

#1175430844685484042

[covert lake]

@marble jetty pls dont promote here

[molten jay]

hi guys

What is the free alternative to the cursor or copilot?

[tough lark]

can i get that documentation link tho

[inland canyon]

where do i download

[desert lion]

Download what

[inland canyon]

the program

that let's you change voice

[desert lion]

Ohh

[inland canyon]

the github is too complicated

[desert lion]

Specify more in #✨│ai-help or #1192011222023950368

Someone will help you faster there

[heavy viper]

Helloo

Is there any ai for notes and everything

[desert lion]

Notion AI 👍

[heavy viper]

Like for any subject?

Is it free??

[desert lion]

Yes Great for organizing all notes, tasks, and projects. AI can summarize your notes, generate outlines, and answer questions from your own notes.

[desert lion]

Yesn't

You get Notion’s base features for free

[heavy viper]

Oh hmm

[polar flax]

basically you can even use chatgpt/gemini for the notetaking purpose, but there are many wrapper applications you could search for, besides notion perhaps

[heavy viper]

I used chatgpt but it's not that great ngl

[polar flax]

because there are better ones like gemini 2.5 pro, claude, etc. but it depends on the purpose

[heavy viper]

Hmm okiee

[main crag]

If anyone want to team to build an AI system please dm!

[desert lion]

<@&1159293140440723499>

[cold iron]

when they say epochs what does that mean?

[short birch]

epochs means that how many loops is gonna train on your dataset

[chilly lake]

or just a note about how many epoch the model was trained

more <> better

[final rivet]

whats the updates way on making ai song covers now?

the prev one isnt working

[hollow field]

is there a choir uvr model?

to isolate/remove choir

[covert lake]

You can check https://docs.aihub.gg and ask in #1192011222023950368 , be sure to not use youtube video tuts tho

[true scroll]

does anyone wanna create an ai with me its going to be huge

[queen jungle]

anybody tried this karaoke model? https://huggingface.co/becruily/bs-roformer-karaoke/tree/main this looks not compatable for previous bs roformer model codes?

[ruby vessel]

where do i find the latest applio version

[severe badger]

hello

how do i get started with ai leads

[modern aspen]

I'm back

Anyways

U guys think it's possible to build a botnet skynet with AI and ml

[plucky abyss]

no

not yet

come back 10 years

[gentle trench]

I think that's a terminator

don't help him

[covert lake]

Hello, in the docs you can find guides and links to the latest version https://docs.aihub.gg/ :)

The current latest one is 3.5.0

[slow ginkgo]

Hi everyone! Is anyone knowledgeable on bolt.diy?

[summer pecan]

AI for notes, yes!

[long pagoda]

hello

[long pagoda]

Notion implemented an AI system

[heavy viper]

[heavy viper]

The thing is I'm looking for one that is free

[odd sundial]

hello

hey

[frigid egret]

hello

[odd sundial]

hi

[spring locust]

did anyone see: "https://www.worldlabs.ai/blog/bigger-better-worlds" — I think it's game-changing

[flat cipher]

Hey,
i am looking for an ai tool that easy to work with and can create for me a workflow diagram so i can add to a presentation. can anyone direct me the right way?

[spring locust]

https://www.eraser.io/diagramgpt

[flat cipher]

thank you i will check it our

[gusty quest]

hello i have an zip file with an ai playboi carti model and i would want to know how to use it and where

[spring locust]

https://chatgpt.com/ / https://grok.com/ — are optimal solutions

[flat cipher]

Did anyone found an easy way to make video clips to your songs? everyway i tried either took me way too long until i gave up or is just a very bad result

[gusty quest]

and like tf should i do with chatgpt like actually

[inner vapor]

what is the best cover app or web?

[polar flax]

you should post a thread in https://discord.com/channels/1159260121998827560/1192011222023950368

[covert lake]

if you're talking about ai covers, it's best you use #1192011222023950368 and check the ai hub docs for help :)

[solar torrent]

Sometimes, chatbot sites like ChatGPT and Google Gemini don't always give you the correct answer you're looking for. The user asked for which progran where "voice model" is supposed to be.

[languid shard]

hello
if only were those tools omnipotent

[solar torrent]

The voice model you have downloaded is likely be RVC. Any RVC program like Applio and the mainline can work. Also, it's best to ask in #✨│ai-help or #1192011222023950368 if you're looking for where to download and use it.

[gentle trench]

That guy is probably talking about using it for realtime

[solar torrent]

Not everything should be "realtime voice changer", you know. He might not mean to look for realtime but something like AI cover, unless he stated more of his needs.

[gentle trench]

it's my assumtion with literally anyone who joins here because ai-help is full of them asking for help with it

[solar torrent]

Nah, you better not judge it like that. It's my duty to help people, not to judge or banish them.

[sterile gale]

I hope the server logo tag will be updated soon

I think this is too old

[languid shard]

Now I get how long it'd take to train a single model. I didn't think it'd take that long 🤣 and Google Colab isn't free either, I guess

[gentle trench]

kaggle is better

30 hours for free users

google colab gives like 2 or something

[languid shard]

200 epochs would take me nearly 3 days

🤣

[gentle trench]

but I'm having issues with kaggle apllio just not working

so use with caution

[gentle trench]

I remember those times

god awful

[languid shard]

I once created an AI imagine through SD 1.5 and also upscaled it. The generation took quite a bit but upscaling to 4k took 4 hours.

yeah, I'm very limited, but I kinda get into 4GB VRAM withut crashing or anything

and fixing a hand in Krita... and "accidentally" selecting a bigger area resulted in crash.

[green violet]

I need an AI voice of Tazoh so bad 🙏

Hio chat how are yall :D

[gentle trench]

You can make one, I believe in you

[green violet]

how

[gentle trench]

Go ask in ai help, I'm not home currently so I can't really help real good

[covert lake]

like the AI Server tag logo?

If we get more boosts we got the server tag logos packs

[covert lake]

check if one exists in #1175430844685484042

[green violet]

I already did, there isn't

[covert lake]

You can search rvc ai voice models at:

• In https://discord.com/channels/1159260121998827560/1163592055830880266 , Do /find with @hidden grotto
• https://weights.com/ (login required)
• https://huggingface.co/models (but watch out cus in hugging face there arent only rvc ai voice models)
• https://voice-models.com/ (just scrapes the huggingface link from the voice-models channel)
• https://thevoicemodels.com/ (for Turkish Models, login required with discord and level 2 on their server)

if there isnt one, you can:

• make it yourself with our docs guides
• Ask a free request in https://discord.com/channels/1159260121998827560/1159289738314919936
• Be aware that we don't allow any paid comms in the server

[hidden grotto]

:wave: @covert lake, How can I help?

Available Commands:
• @weights find <query> or /find <query> - Search for RVC Voice Models

[covert lake]

gave you all possible ways to find models

and also what to do if no one ever made it :)

[green violet]

Thx :'D

[green violet]

How do i request a model im too stupid

[languid shard]

Guys, how many epochs and how big of a dataset should I prepare? I've read some resources in AI HUB and some videos but opinions varies;

• 40 min. for a realistic voice. a minimum of 10 min. is recommended
• Also what should be the length of the clips? There is also high variance from 5 to 20s
• clear voice without lip smacking and other sounds
  Is there anything else? I plan to do this in Audacity (so good for learning some new functions)

[gentle trench]

Most likely it'll not be seen by anyone who actually makes models, I'd recommend learning how to make them

[green violet]

wah

[gentle trench]

I will sometimes on rare occasions make models for my close friends but not often

[green violet]

I'll see what to do :p

[gentle trench]

Yeahhhh there's a section in the server for you to request models but nobody ever looks there tbh

[covert lake]

click a new post

[green violet]

I mean i know that but the thing idk how to request it

[covert lake]

You just put the character Name and voice actor, along with the Language

[green violet]

what if its a youtuber-

[covert lake]

The youtuber name and language

[green violet]

oh

[green violet]

Small lil question 🥹

[covert lake]

do you want to ask another smal lil question or was just making a comment about your question?

also dw it's fine

[green violet]

NO LIKE

I WANT TO ASK

YOUR PFP IS

MURDER DRONES???

😭

[covert lake]

https://tenor.com/view/murder-drones-serial-designation-n-murder-drones-n-knocked-tf-out-rebooting-gif-17693361778907942337

Yupp

[green violet]

OMG AAA

[covert lake]

I recently found out about the show less than a month ago lmao

[green violet]

I don't know anyone that knows that youtube serie okay 🙏

[green violet]

I KNOW IT SINCE IT CAME OUT

KJFAJKSFJAN

[covert lake]

It's finee dw, I like talking about things I like

[covert lake]

holy 😭

must be pain to wait this much time for episodes

well i'm doing the same for TADC

[green violet]

real

[green violet]

me too hehaehahsd

[green violet]

GKAJSFNM which you think will abstract first?!?!?!!?!??! >:)

[covert lake]

I can see that you're also pretty shy, dw I wasn't going to eat u if it wasn't murder drones 😭

honestly I can't blame you

I joined this server july 2023

I started chatting and becoming staff in september 2023

and boom now im junior admin

if i never sent some silly messages at first you'd prob not meet me there rn

[covert lake]

it feels like everyone wants to abstract tbh 😭

Caine is having very big glitches recently tho

[green violet]

LOL THX TO NOTICE IT I THINK IDK

[covert lake]

the first post credit scene is caine glitching, which may be foreshadowing

[green violet]

For me that he breaks down in some moment

he already screamed like 8000 times at zooble

[covert lake]

https://tenor.com/view/sounds-a-little-convoluted-you-look-convoluted-tadc-zooble-caine-gif-10600043665045312823

fr lol

hey btw is it fine if we switch to #✦│chat ? i know those shows are about ai technically but I just don't wanna clutter here lol

[green violet]

lol yes yes sure :'D

[green violet]

I joined recently

[tired scaffold]

Hello

World

.

balls

[main yacht]

Do the voices only work for singing?

[green violet]

they also work for talking

[main yacht]

Hahaa bettt

[green violet]

But i think you need to search specific voices for it

[main yacht]

I wish they could he added to voice mod tho😭

[green violet]

because singing only models may not work properly

[tired scaffold]

what even is this place? is this a specific product or just an AI server?

[green violet]

I mean

[tired scaffold]

can i speak to the manager??????/

[green violet]

the name literaly says it

[green violet]

I AM THE MANAGER /J

[main yacht]

Yea i wanted a good do not redeem one

[tired scaffold]

ok smarty pants, i dont have time to read stuffs all the time

[green violet]

Okay sorry-

[tired scaffold]

cool sweet

[green violet]

My bad-

[tired scaffold]

shame on you....

hehe lol

i got banned from the singularity discord

like a few months ago

because i foretold the inevitable truth

[gentle trench]

voice mod is dookie compared to the real time voice changer

any realtime vc plus fl studio beats voicemod because it can be all done for free

only pay voicemod for the soundboard

[solar torrent]

Unlikely. W-Okada and Vonovox are better than Voicemod, while Voicemod is for filtering your voice but not RVC.

[solar torrent]

So isn't this server supposed to be all about Fortnite? Lol.

[main yacht]

Gotcha whats the best one that has access to alot of voices and works good no lag and stuff

Free or not doesn’t matter

[gentle trench]

voice changer right?

[gentle trench]

https://cdn.discordapp.com/attachments/1360063328730480735/1387975466769584229/togif.gif

[polar flax]

wut

https://cdn.discordapp.com/attachments/427903255096459294/1417769418842181632/IMG_9975.jpg?ex=68cbb02a&is=68ca5eaa&hm=da6820393a8b2de1b1495791e5a23b65b3ada6c90e671475422e99142d1314f4&

[solar torrent]

For W-Okada realtime voice changer or other AI program, go to #✨│ai-help or #1192011222023950368 and explain about your needs there. For voice models, there are in #1175430844685484042.

[static sequoia]

Is there any difference in voice quality between Vonovox and W-okada?

[graceful flame]

Yeah, making decent video clips for songs can get frustrating if you’re doing it all manually. It takes a lot of time for creating and editing part. you can start with an AI tool( AI Video Generator) and then give touchup with a lightweight editor. Gets you 80% of the way there without the headache.

[terse vapor]

whose

[autumn wharf]

hi

[gentle trench]

Vonovox makes human models sound more realistic

Less artifacts and stuff

[languid shard]

Hi. I'd like to ask if there is a way to make a model recognize laughing and so on if I create a separate folder of shorter wav files and other things, to further expand a model. Also I read about the further training of a model but if a model has .pth file, index and config, can it be trained further if the training is finished? (I read that it can be, but just to make sure) 🙏

• in general, what files I need in order to improve an already existing model (to keep it short)

[odd sundial]

did u generate ur pfp or you got it from google

[languid shard]

I think it was a website, like 2 years ago when I first stumbled upon AI.

[odd sundial]

im looking for a server that has a lot of ai pfp users on it

[languid shard]

I think it was generated on a nightcafe website

[odd sundial]

I forgot that shit still exists

[languid shard]

it was fun to play around with it, I also used a local comfyUI GUI and just piped nodes and such

[odd sundial]

I just use chatgpt tbh to do a lot of artwork but im still tryna make a character model of mine, im just so trash at it, do yk anyone who can do this for me

[languid shard]

hmm, I wanted to try creating my own model, but quickly gave up on the images because it needs a lot more powerful GPU, but I think AI voice model should be feasible

I just google and went through docs and youtube + used chatGPT as a helping hand for search

[limber nexus]

So i need ideas to add to my AI that i made any suggestions? She can already speak and search google.

[gentle trench]

That sounds cool

You should make her have the ability to play games

On a browser or something

[main yacht]

I tried using vonovox but it said path could not be found

And when I press a button like it indicates me to do cmd closes

[gentle trench]

hm

what gpu do u have

[main yacht]

4090

I have a pic of my error

[covert lake]

Could you use #1192011222023950368 or #✨│ai-help to not clutter the chat please?

[main yacht]

Yep sorry

[limber nexus]

Yeah that was something i was thinking about i just need to figure a out a simple game, like Plants Vs zombies, or Five nights at freddys, you know a game that doesnt have 3D movement for now at least

[gentle trench]

I love pvz and fnaf

I think an ai that could recognize what plants do what and how to counter the specific zombies in each level would be awsome

[limber nexus]

Yeah i think that would work with the biggest thing being i have to get a photo of each plant for the image recognition, but i think i can do that thanks for your suggestion

Also have you made a Patrick star RVC model cause i remember that name before

[gentle trench]

I have yea

this right herehttps://discord.com/channels/1159260121998827560/1376987620822880256

most recent one I made of him

[limber nexus]

i knew it! i used that model a few times and i love it cause it sounds like season 1 patrick, and i dont know about you but season one spongebob is the best

[gentle trench]

dude the first 3 seasons

https://tenor.com/view/imspongbesa-gif-14794359968060830097

amazing

[limber nexus]

Yeah they were Peak, then the next best seasons were up until like, hmm say season 7ish

[gentle trench]

the models I made of all the characters are based off games so that's why Spongebob doesn't have as high pitch of a voice as usual

[limber nexus]

ohh okay that makes sense, also i dont train any models but it seems to me that its difficult even for AI to get spongebobs voice right, am i crazy or is that the case?

[gentle trench]

tbh I feel like now it could get it right if there was enough audio

back then it was harder

[limber nexus]

makes sense considering how we got things nowadays, so im curious how long does it take to train a RVC model on whatever GPU you have?

[languid shard]

it highly depends on your specs

[chilly lake]

32min set, fp32, 4070 ti super

[languid shard]

set? not epoch?

[chilly lake]

i mean dataset size

130epochs done in an hour, 200e was a bit overtrained already

[languid shard]

I have a question, what is the recommended size of steps and epochs for the first time wiht a dataset as large as 40 min?

I'll aim for a lot better Nvidia GPU then, that's a huge difference.

[chilly lake]

(40 min * 60s / 3s chunks) * 1.1 = number of chunks / batch size = number of steps per epoch

[gentle trench]

scary math...

[languid shard]

Where did you get the formula and what are those chunks? thanks

[chilly lake]

so with batch 8 it is like 110s/epoch

[limber nexus]

Look equations like that are the ones that make me wonder how anyone rememebers them

[chilly lake]

what is there to remember? the dataset gets sliced into 3s chunks with 0.3s (10%) overlap

40 min is 880 chunks

[languid shard]

so many opinions on this I have like 5s min. per chunk

[chilly lake]

fp16 is about 2x faster than fp32.. much faster on T4 (colab/kaggle) or any other AI specific professional gpu

5s you waste about 90% of the dataset

[languid shard]

that's a Japanese

I'll be editing forever

[limber nexus]

yeah well i dont remember things like that cause its difficult for me, the only equation i remember is: P = V × I

[chilly lake]

chuck size affects the long term dependency, but the generator is trained on just 0.35s

there's no need to manually edit, just feed one big file and it will slice it

[languid shard]

I haven't checked anything that'd do this process automatically except URV or what is it called. Is there any tool taht'd make it automatic?

[gentle trench]

https://tenor.com/view/patrick-star-overthinking-homework-tired-brain-brain-on-fire-gif-16625827878902407192

[chilly lake]

preprocess slices audio automatically

[languid shard]

so I just click on preprocess in Applio and this will do?

[chilly lake]

just remove long silences in audacity

[languid shard]

yeah, I'm going manually because I couldn't get the right truncate settings in Audacity, it sliced off some words too.

[chilly lake]

and read the preprocess options

[limber nexus]

i forgot i have audacity on my PC till now thanks for reminding me

[chilly lake]

this does not truncate words

[languid shard]

yes, it can happen, 1 wav was ok, the 2nd wav had a few words sliced

[gentle trench]

btw if u wanna look at everything I made that was posted publically just look here at all my models since not everything is in the discord serverhttps://www.weights.com/user/your_local_worm

[limber nexus]

Thanks ill look at those later, in fact i used your patrick model to make a cover of Hello Kitty by avril lavigne and it turned out to be the most cursed thing ever

[gentle trench]

XD

is it on youtube?

[glad nebula]

the cool part about mainline's auto slicer is the fact you dont need to truncate the silence or remove long noisy silence manually, or at least it was able to correctly slice p225 files without creating slices of pure noise

[limber nexus]

no it isnt i was thinking about it but i can share it here if you want to hear it

[gentle trench]

if it has music u can't share here but u could in dms

copyright with music and covers here is strict

that's why samples always only have the voices

[glad nebula]

discord hates instrumentals for no reason

[limber nexus]

ahh okay then, sorry im new to this server but if you want to hear it let me know

[gentle trench]

I meant the server rules

we don't wanna get sent to the void again

[glad nebula]

they can also ban u even if you share them in dms

lol

i mean not weights, but the annoying discord bot

[limber nexus]

seriously? thats just ridiculous

[glad nebula]

ai based moderation basically

[limber nexus]

from my experience AI has a hard time detecting AI

[gentle trench]

bro whahttt

[glad nebula]

seoul told us last time he got banned because copyright nonsense despite him never posting any instrumentals in a server

[limber nexus]

so why are the're discord bots that can play copyrighted music then?

[glad nebula]

so ye its possible

[gentle trench]

hypocrites

[limber nexus]

honestly things like this need to have something done with them, like if i use spotify that is distributing free music if i dont pay and that should be illegal right?

[gentle trench]

fr

tik tok has tons of copright music

tons of platforms

youtube allows ai covers

[limber nexus]

yeah and ive used Python librarys that download music from youtube even just me downloading it should be illegal but it isnt

[gentle trench]

smh

[chilly lake]

except it ends up atrociously long

[gentle trench]

imagine getting banned over an mp3 file

[limber nexus]

yeah its the same thing like going to jail for something dumb like i saw a girl wouldnt get out of her dads car and went to jail for it, so its like why you in jail? i didnt get out of a car. its dumb

[plucky abyss]

hi

[plucky abyss]

wth how do you go to jail for that?

[limber nexus]

so it was a youtube video i found, it was a Girl that was kicked out of a mental hospital and wouldnt get out of her dads car so she was arrested for resisting and tresspassing in her dads car, i know what a dumb way to go to jail

sorry if that wasnt very informitive but thats about as much that i can think of that answers your question

[dull jewel]

hi

[limber nexus]

hi

[gentle trench]

hi

[dull jewel]

how are you?

[limber nexus]

me? im good how about you?

[dull jewel]

too, going to sleep in 10 mins

[plucky abyss]

that is very weird. Don't think i've ever seen anybody be arrested for not getting out of a car (that's not like a commercial vehicle)

[limber nexus]

oh okay, well have a good rest then

[dull jewel]

thank you

[limber nexus]

well me neither but the daughter shouldnt have been out on the streets, she was even trying to convince the cops that her father had early onset dementia when he clearly didnt

[short birch]

gn

[plucky abyss]

boy the mentally unstable trying to vonince authorities that their caretaker is nmentally unstable lmao

[limber nexus]

i know!, and at one point she said that she was being held against her will in the car, so one of the cops said "Great get out" and she just said that it was unrelated

[plucky abyss]

lmao

[limber nexus]

Yeah, So anyway forgot to ask how are you?

[plucky abyss]

Doing okay i guess all things considered. How about you ?

[limber nexus]

Not horrible just juggling a bunch of things at once i guess

[plucky abyss]

fun. me too lmao\

[limber nexus]

i also have to work on fixing my AI which is being a Pain, any projects you working on?

[plucky abyss]

yeah, working on my Funding SaaS platform, handling some clients special rquests and started doing some work with a fried to try to help him technologial . they're on Zoho and long started growing out of it

what's your AI doing ?

[limber nexus]

Oh so i dont have a real GPU yet so im using Groq API and she keeps getting a unknown response error, it started once i changed the system prompt and im still working on fixing it

[plucky abyss]

ahh

[limber nexus]

Yeah and i mean its not a huge deal, cause once i do get a GPU everything should be much smoother running on my PC localy

anyway whats up with your SaaS platform?

[plucky abyss]

its a funding SaaS Platform.; iono what more to say

been working on it for a few years

[limber nexus]

oh okay then just curious y'know?

[plucky abyss]

what country are you in?

[limber nexus]

North America Why?

[plucky abyss]

you ever seen on tv advertisements for "small business funding" (companies like OnDeck / Kabbage) ?

[limber nexus]

no not really, the only ads i get are on youtube

[plucky abyss]

ok

well my platform ios a whitelabel platform that funders can use to underwrite and fund clients (can do pretty mnuch anything) but i primarily focus on small business funders (less regulated than consumer stuff)

[limber nexus]

oh wow, thats cool, more than what i could ever do. how was writing in Java script? i hated trying to use it

[plucky abyss]

i use typescript which transpiles to Javascript (but onlyfor the front end).. i do not like it ..

lmao

but it is a necessary evil

typescript makes some aspects of javascript more manageable but still can't get away from the "javscript-ness"

lm;ao

[limber nexus]

yeah for any javascript i do i just use ChatGPT cause i am not dealing with that mess of a programming language

[plucky abyss]

my project is.. way too big for AI

lol

just the TS alone over 2MB of TS code

250 files

and that's not the UX

thats just everything that the UX uses .

[limber nexus]

Wow that is massive, at least compared to anything ive done, good job and good luck on working on that

[plucky abyss]

thanks, yeah its been a long.. arduous journey

[limber nexus]

wish i could say the same, the biggest project ive made is my AI which is only like 450 lines of code in python, which i still have to add more to

[plucky abyss]

yeah well keep at it 🙂

i've been doing this for 25 years 😄

[limber nexus]

thanks i will, also that is a insanely long time dont you ever get bored?

[plucky abyss]

This has kind of been my passion for the last 25 years so no, not really

I have been suffering from burnout and a bit of depression for the last year+

Kind of have been evaluating if it is time to start pivoting to something else

[limber nexus]

oh well i hope you will get things straitened out and if needed find something else you like to do for a job. i would say i know about depression but i only have ADHD so i cant say

[plucky abyss]

yeah

its just a part of life

[limber nexus]

Yeah i mean my dad has depression and he dropped it like out of nowhere and i didnt even know! so i know you can get through it just keep going

[limber nexus]

Well sorry but ive got to go, if you want to talk more later just let me know, but for now bye👋

[plucky abyss]

laters. have a good one

[limber nexus]

you too

[languid shard]

Hello today again. I havea quesiton about pre-trainedm odels. Which one should I use for Japanese? I currently have Titan Medium 40kHz

bye Hlubby

[minor blade]

Avoid using any pretrain that isn't the OG default one

Also, for further help go to the #✨│ai-help channel

[vagrant void]

How come models are still frequently uploaded here? is there any training colab that still works today?

[void relic]

Hi Hlubby

[twin valley]

Which version of KLM would you suggest for a male singer ? ☺️

[limber nexus]

Hi spiderman

[solar torrent]

For known RVC training Colab notebooks, see #🤖│bots message, or ask in #✨│ai-help or #1192011222023950368.

[gentle trench]

Because voice models are super sick and awesome

[granite bloom]

hi

[solar torrent]

RVC voice models are funny. I used some of them to sing "BBL Drizzy".

[solar torrent]

What help?

[languid shard]

kinda late, I used Snowie 3.1, hopefully it isn't bad, I read about it on APplio website

[slate saffron]

Hi chat

[silent bolt]

Is VCClient legit? can I be sure it isn't a trojan?

[covert lake]

replied to you there #✦│chat message

[marsh venture]

Guys, is there a model for this voice?

https://www.youtube.com/watch?v=_y9MpNcAitQ

[queen kernel]

If you know the name you can search in #🔍│find-models

[marsh venture]

say pls

i wanna use this voice

[marsh venture]

ok

[queen kernel]

I don't know what this voice is called

[marsh venture]

ok bro

[covert lake]

https://www.reddit.com/r/StableDiffusion/s/oHt5eqSZay
@stark scarab I feel like China banning Nvidia gpus can end in either:

• new china GPUs with proprietary CUDA-Like
• new CUDA-Like software for every GPUs
• rip china ai (99.999% never happening)

👀

[gentle trench]

Damn they banned the only good gpus

[covert lake]

it may be a good thing if they do smt like Zluda but better

[gentle trench]

Like what

What ancient god is that

[polar flax]

I thought the ban doesn't include consumer gpus lower than 4090/5090

[covert lake]

https://github.com/vosen/ZLUDA

[covert lake]

https://www.cnbc.com/2025/09/17/nvidia-ceo-disappointed-after-reports-china-has-banned-its-ai-chips.html

[plucky abyss]

lol

[spark depot]

What's the most human like AI?

[chilly lake]

that's original zluda, but they are making some progress again